Spatially explicit analysis of estuarine habitat for juvenile winter flounder: combining generalized additive models and geographic information systems
Aw. Stoner et al., Spatially explicit analysis of estuarine habitat for juvenile winter flounder: combining generalized additive models and geographic information systems, MAR ECOL-PR, 213, 2001, pp. 253-271
Quasisynoptic seasonal beam trawl surveys for age-0 winter flounder Pseudop
leuronectes americanus were conducted in the Navesink River/Sandy Hook Bay
estuarine system (NSHES), New Jersey, from 1996 to 1998, to develop spatial
ly explicit models of habitat association. Relationships between environmen
tal parameters and fish distribution were distinctly nonlinear and multivar
iate. Logistic generalized additive models (GAMs) revealed that the distrib
ution of newly settled flounder (< 25 mm total length) in spring collection
s was associated with low temperature and high sediment organic content, pl
acing them in deep, depositional environments. Fish 25 to 55 mm total lengt
h were associated with high sediment organics, shallow depth (<3 m), and sa
linity near 20 ppt. The largest age-0 fish (56 to 138 mm) were associated w
ith shallow depths (< 2 m), temperature near 22 degreesC, and presence of m
acroalgae. Abundance of prey organisms contributed significantly to the GAM
for fish 25 to 55 mm total length, but not for fish > 55 mm. Independent t
est collections and environmental measurements made at 12 new sites in NSHE
S during 1999 showed that the GAMs had good predictive capability for juven
ile flounder, and new GAMs developed for the test set demonstrated the robu
stness of the original models. Maps for the probability of capturing fish o
f particular sizes were produced by integrating GAMs with maps of environme
ntal data in a geographic information system (GIS). These plots revealed 2
important centers of settlement in the system, probably related to hydrogra
phic conditions, and the fact that nursery locations shift rapidly with fis
h size during the first year of life. It was also apparent that nursery hab
itats are dynamic-expanding, contracting, and shifting position with change
s in key environmental variables.